‘Productivity and cost savings matter, but they are only the first layer’: Deloitte expert
Canadian organizations are seeing strong returns from artificial intelligence (AI) but may be limiting the technology's impact on the workforce by measuring only its most immediate, easily tracked benefits, according to a recent report from Deloitte Canada.
Overall, 88% of senior leaders at Canadian organizations are confident they can measure AI return on investment, and 90% reported positive productivity impacts over the past 12 months.
Despite that confidence, Deloitte Canada found fewer leaders cited broader outcomes, such as revenue growth or risk reduction, as evidence of AI's value, compared with those pointing to productivity gains and cost savings.
"Most organizations are still measuring the easiest value, not the most important value," said Nihar Dalmia, partner in AI & data at Deloitte Canada, in the release. "The leaders seeing more differentiated outcomes are expanding how AI value is measured to include revenue growth, faster decision-making, better customer experiences and new ways of working."
Organizations take incremental approach
Deloitte Canada also found that operations ranks as the top function for realized AI value today, cited by 57% of respondents, suggesting many employers are using AI to improve existing processes rather than redesigning work.
"Most leaders are still taking an incremental approach to AI because it's the path of least resistance," said Jas Jaaj, global AI markets, ecosystems & alliances leader at Deloitte Canada, in the release.
Jaaj said in the release that "the step change comes when organizations expand their focus beyond efficiency and rethink how work gets done end-to-end."
Confidence high, but oversight gaps remain
Nearly all (95%) leaders expressed confidence in their organization's AI compliance, and 94% in internally developed AI systems, according to Deloitte’s survey of 300 senior leaders at Canadian organizations, most with more than $100 million in annual revenue, conducted in March. Nearly two-thirds (64%) estimated at least 11% of AI activity at their organization occurs outside approved or governed tools.
"What stands out is that trust itself isn't really the issue anymore," Dalmia said in the release. "The bigger challenge is visibility and control."
The survey also found 96% of leaders said it is important that AI technologies be developed, hosted and governed under Canadian law.
Data quality cited as top barrier
Deloitte Canada identified data quality, cited by 59% of respondents, and legacy system integration, cited by 51%, as the most common barriers to further AI value. Better data was named by 45% as the biggest accelerator of stronger outcomes.
Leaders are continuing to invest in upskilling, with 86% planning to increase AI-related training, and 92% reporting employees view AI's impact positively.
"The leaders who are unlocking value aren't waiting for perfect data," Dalmia said in the release, adding that those leaders are disciplined about focusing on individual use cases while also improving the data and systems tied to their organization's biggest priorities.
One in three Canadian workers admit to faking their AI abilities at work, according to a previous report.
Here’s how employers can successfully measure AI ROI, according to multiple sources:
|
Instruction to Employers |
Supporting Data |
Source |
|
Do not rely on token consumption or usage volume as a proxy for AI value; measure how work and outcomes are changing instead. |
Token consumption has become a default proxy for AI engagement, but it measures activity rather than value, according to Carolyn Hamer, partner, human capital at Deloitte. |
Deloitte expert |
|
Treat pilot results with rigour before scaling, since most AI pilots do not yet show a measurable financial return. |
Roughly 95% of generative AI pilots reviewed showed no measurable effect on profit and loss; the finding is preliminary and not peer-reviewed. |
MIT Project NANDA, "The GenAI Divide: State of AI in Business 2025," reported by Fortune |
|
Redesign workflows around AI rather than layering it onto existing processes. |
Fundamental workflow redesign is the strongest predictor of earnings impact among organizations surveyed; only a minority of AI adopters have undertaken it. |
McKinsey & Company, "The state of AI in 2025: Agents, innovation, and transformation," Nov. 5, 2025 |
|
Expand ROI measurement beyond productivity gains and cost savings to include revenue growth, decision-making speed and new ways of working. |
88% of leaders are confident they can measure AI ROI, but the most common indicators cited remain productivity gains and cost savings rather than broader business outcomes. |
Deloitte Canada survey, released July 28, 2026 |
|
Move past isolated pilots toward governed, enterprise-wide adoption before expecting returns. |
Nearly half (46%) of Canadian business leaders are experimenting with AI without achieving meaningful ROI; only 18% are actively embedding AI into workflows and operations. |
BDO Canada, "AI Vision Report: Past the pilot to the agentic future of work," June 25, 2026 |
|
Benchmark integration progress against comparable organizations to identify readiness gaps. |
Canadian companies continue to trail U.S. counterparts in AI integration, transformation and investment returns, despite high levels of investment. |
RSM Canada, "Middle Market AI Survey 2026: U.S. and Canada," July 21, 2026 |
|
Give employees explicit guidance on where and how to apply AI, rather than assuming enthusiasm will translate into use. |
65% of workers are excited to use AI, but only 42% know how to identify where it can improve their work. |
Gartner research, reported by HRD Canada, Oct. 28, 2025 |
|
Prioritize closing workforce skills gaps, since a lack of skills is the primary obstacle standing between AI use and measurable returns. |
70% of organizations say AI is delivering meaningful business value, but only 3% have achieved measurable returns on their AI investments; a workforce skills gap was cited as the primary obstacle. |
KPMG research, reported by Canadian HR Reporter, May 6, 2026 |
|
Pair AI rollout with structured, funded training programs rather than leaving skill-building to employees. |
84% of workers and 64% of managers received no AI training in the past year, even though 63% of employees view developing AI skills as important. |
Dayforce research, reported by Canadian HR Reporter, Oct. 6, 2025 |
The best way to get the utmost benefit out of AI is to deploy it deeply across operations rather than confining it to pilot projects, according to a recent global report.
